Multi-site solar power forecasting using gradient boosted regression trees

Persson, C.; Bacher, Peder; Shiga, Tohru; Madsen, Henrik

doi:10.1016/j.solener.2017.04.066

Cited by 316 publications

(126 citation statements)

References 14 publications

Supporting

Mentioning

123

Contrasting

Unclassified

Order By: Relevance

“…Boosting regression trees incorporate the strengths of both regression trees, handling various types of predictor variables and accommodating missing data, thus boosting and improving predictive performance by combining many simple learners (Elith et al, 2008). Gradient boosting regression trees (GBRT) is one of the most widely used boosting regression trees and has been recognized as a powerful and successful ML technique in a wide range of practical applications (Natekin and Knoll, 2013;Persson et al, 2017). In GBRT, a gradient descent algorithm is used to minimize the squared error loss function.…”

Section: (B) Boosting: Gradient Boosting Regression Trees (Gbrt)mentioning

confidence: 99%

Predicting wind pressures around circular cylinders using machine learning techniques

Kwok

2020

Journal of Wind Engineering and Industrial Aerodynamics

104

View full text Add to dashboard Cite

Numerous studies have been carried out to measure wind pressures around circular cylinders since the early 20 th century due to its engineering significance. Consequently, a large amount of wind pressure data sets have accumulated, which presents an excellent opportunity for using machine learning (ML) techniques to train models to predict wind pressures around circular cylinders. Wind pressures around smooth circular cylinders are a function of mainly the Reynolds number (Re), turbulence intensity (Ti) of the incident wind, and circumferential angle of the cylinder. Considering these three parameters as the inputs, this study trained two ML models to predict mean and fluctuating pressures respectively. Three machine learning algorithms including decision tree regressor, random forest, and gradient boosting regression trees (GBRT) were tested. The GBRT models exhibited the best performance for predicting both mean and fluctuating pressures, and they are capable of making accurate predictions for Re ranging from 10 4 to 10 6 and Ti ranging from 0% to 15%. It is believed that the GBRT models provide very efficient and economical alternative to traditional wind tunnel tests and computational fluid dynamic simulations for determining wind pressures around smooth circular cylinders within the studied Re and Ti range.

show abstract

Section: (B) Boosting: Gradient Boosting Regression Trees (Gbrt)mentioning

confidence: 99%

Predicting wind pressures around circular cylinders using machine learning techniques

Kwok

2020

Journal of Wind Engineering and Industrial Aerodynamics

104

View full text Add to dashboard Cite

show abstract

“…In this work, the forecasting tasks are treated as regression problems and machine learning regression algorithms on scikit-learn package in Python are adopted to build the models using the Gradient boosted regression trees (GBRT) algorithm. The GBRT algorithm has a superior advantage of not requiring complex data pre-processing of dimension transformations or reduction and does not suffer any loss of input variable interpretation [53]. The significant feature of the accurate implementation of the GBRT algorithm is the parameter α gbr called the learning rates.…”

Section: Gradient Boosted Regression Trees (Gbrt) Model For Time-aheamentioning

confidence: 99%

Multi-Objective Optimal Capacity Planning for 100% Renewable Energy-Based Microgrid Incorporating Cost of Demand-Side Flexibility Management

et al. 2019

View full text Add to dashboard Cite

The need for energy and environmental sustainability has spurred investments in renewable energy technologies worldwide. However, the flexibility needs of the power system have increased due to the intermittent nature of the energy sources. This paper investigates the prospects of interlinking short-term flexibility value into long-term capacity planning towards achieving a microgrid with a high renewable energy fraction. Demand Response Programs (DRP) based on critical peak and time-ahead dynamic pricing are compared for effective demand-side flexibility management. The system components include PV, wind, and energy storages (ESS), and several optimal component-sizing scenarios are evaluated and compared using two different ESSs without and with the inclusion of DRP. To achieve this, a multi-objective problem which involves the simultaneous minimization of the loss of power supply probability (LPSP) index and total life-cycle costs is solved under each scenario to investigate the most cost-effective microgrid planning approach. The time-ahead resource forecast for DRP was implemented using the scikit-learn package in Python, and the optimization problems are solved using the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm in MATLAB R . From the results, the inclusion of forecast-based DRP and PHES resulted in significant investment cost savings due to reduced system component sizing.Keywords: demand response program (DRP); photovoltaic system (PV); pumped heat energy storage (PHES); critical peak pricing (CPP) DRP; time-ahead dynamic pricing (TADP) DRP; loss of power supply probability (LPSP); energy storage system (ESS); Multi-Objective Particle Swarm Optimization (MOPSO)

show abstract

“…The work proposed in (Barzin, 2016) suggests that the use of gradient boosted regression trees can be a valid solution for multi-site solar power forecasting. In (Persson, 2017) solar forecasting is used as basis for a price-based control mechanism for PV. In (Pedro and Coimbra, 2012) a comparison on several forecasting techniques to predict solar power at a photovoltaic power plant in California is presented.…”

Section: Solar Forecastmentioning

confidence: 99%